CaiWuXiTong/app/Business/JstOrderLabelImporter.php

<?php

namespace App\Business;

use Illuminate\Database\DatabaseManager;
use InvalidArgumentException;
use RuntimeException;

/**
 * 聚水潭订单标签明细导入（拆标签写入 jst_order_labels）。
 *
 * 一个标签一行：一单有 N 个标签就写 N 行；订单不去重、标签不汇总。
 * 无标签的订单保留一行，标签记为「（无标签）」，方便与订单总数核对。
 *
 * 两个数据源：
 *  - 共享镜像 ops_jst_orders（常规，覆盖镜像有数据的区间）；
 *  - 聚水潭「导出订单」Excel（镜像没有的历史区间，例如 2026-03，只能靠财务导出补）。
 *
 * 卖家备注包含「SSDD」时同时生成 SSDD 标签。卖家备注不是聚水潭系统标签，
 * 但种菜判断需要把这类订单纳入标签明细。
 */
class JstOrderLabelImporter
{
    public const NO_LABEL = '（无标签）';

    public const SELLER_REMARK_LABEL = 'SSDD';

    private const READ_CHUNK = 2000;

    private const INSERT_CHUNK = 2000;

    /** 聚水潭导出文件里用到的表头。 */
    private const EXPORT_COLUMNS = ['线上订单号', '店铺名称', '下单时间', '标签'];

    private const EXPORT_OPTIONAL_COLUMNS = ['卖家备注'];

    public function __construct(
        private readonly DatabaseManager $databases,
        private readonly ?XlsxSheetReader $exportReader = null,
        private readonly ?JstOrderLabelItemCodes $itemCodes = null,
    ) {}

    /**
     * @return array{from:string,to:string,orders:int,rows:int,no_label_orders:int,labels:int,seconds:float}
     */
    public function import(string $from, string $to, int $companyId = 1, bool $replace = false, string $source = 'jst-order-label-import'): array
    {
        $from = $this->normalizeDate($from, '起始日期');
        $to = $this->normalizeDate($to, '结束日期');
        if ($from > $to) {
            throw new InvalidArgumentException('起始日期不能晚于结束日期。');
        }

        $start = $from . ' 00:00:00';
        $end = date('Y-m-d', strtotime($to . ' +1 day')) . ' 00:00:00';

        $mirror = $this->databases->connection(config('upstream_mirrors.connection', 'business_mirror'));
        if (! $mirror->getSchemaBuilder()->hasTable('ops_jst_orders')) {
            throw new RuntimeException('聚水潭订单镜像尚未建立（ops_jst_orders）。');
        }

        $target = $this->databases->connection();
        $startedAt = microtime(true);
        $itemCodes = $this->itemCodes ?? new JstOrderLabelItemCodes();
        $hasItemTable = $mirror->getSchemaBuilder()->hasTable('ops_jst_order_item');

        if ($replace) {
            $target->table('jst_order_labels')
                ->where('company_id', $companyId)
                ->where('ordered_at', '>=', $start)
                ->where('ordered_at', '<', $end)
                ->delete();
            app(PlantingOrderLabelLookup::class)->forget($companyId);
        }

        $now = now();
        $orders = 0;
        $rows = 0;
        $noLabelOrders = 0;
        $buffer = [];

        $flush = function () use (&$buffer, &$rows, $target, $companyId): void {
            foreach (array_chunk($buffer, self::INSERT_CHUNK) as $chunk) {
                $target->table('jst_order_labels')->insert($chunk);
                app(PlantingOrderLabelLookup::class)->forget($companyId);
                $rows += count($chunk);
            }
            $buffer = [];
        };

        $mirror->table('ops_jst_orders')
            ->where('order_date', '>=', $start)
            ->where('order_date', '<', $end)
            ->orderBy('Id')
            ->chunkById(self::READ_CHUNK, function ($records) use (&$buffer, &$orders, &$noLabelOrders, $flush, $companyId, $source, $now, $itemCodes, $hasItemTable, $mirror): void {
                // 货号：同批订单的商品编码（镜像商品明细，按 o_id 取），导入时一并写入
                $orderCodes = $hasItemTable
                    ? $itemCodes->forOrderIds($mirror, 'ops_jst_order_item', $records->pluck('o_id')->all())
                    : [];

                foreach ($records as $record) {
                    $orders++;
                    $orderedAt = $record->order_date === null ? null : substr((string) $record->order_date, 0, 19);
                    if ($orderedAt === null || $orderedAt === '') {
                        continue;
                    }

                    $labels = $this->labelsForRecord(
                        (string) ($record->labels ?? ''),
                        (string) ($record->remark ?? ''),
                    );
                    if ($labels === []) {
                        $labels = [self::NO_LABEL];
                        $noLabelOrders++;
                    }

                    $itemCode = $orderCodes[(int) $record->o_id] ?? null;

                    foreach ($labels as $label) {
                        $buffer[] = [
                            'company_id' => $companyId,
                            'order_no' => mb_substr((string) $record->so_id, 0, 100),
                            'shop_name' => mb_substr((string) ($record->shop_name ?? ''), 0, 191),
                            'ordered_at' => $orderedAt,
                            'label' => $label,
                            'item_code' => $itemCode,
                            'created_from' => $source,
                            'created_at' => $now,
                            'updated_at' => $now,
                        ];
                    }

                    if (count($buffer) >= self::INSERT_CHUNK * 2) {
                        $flush();
                    }
                }
            }, 'Id');

        $flush();

        $labels = $target->table('jst_order_labels')
            ->where('company_id', $companyId)
            ->where('ordered_at', '>=', $start)
            ->where('ordered_at', '<', $end)
            ->distinct()
            ->count('label');

        // 导入会改变筛选选项，重新预热缓存，避免页面下次打开时做几秒的 distinct 扫描。
        app(JstOrderLabelOptions::class)->forget($companyId);
        app(JstOrderLabelOptions::class)->warm($companyId);

        return [
            'from' => $from,
            'to' => $to,
            'orders' => $orders,
            'rows' => $rows,
            'no_label_orders' => $noLabelOrders,
            'labels' => (int) $labels,
            'seconds' => round(microtime(true) - $startedAt, 1),
        ];
    }

    /**
     * 从聚水潭「导出订单」Excel 导入。
     *
     * 导出文件是一单多行（商品明细），且同一单里可能只有部分明细行带标签：
     * 第一遍先挑出真正有标签的订单号，第二遍再按「线上订单号 + 标签」去重拆行写入，
     * 只对整单没标签的订单写「（无标签）」。同时兼容只勾 4 列导出的极简表头。
     *
     * 传 $from/$to 时会顺带按窗口过滤；带 $replace 时必须给窗口，
     * 以便在写入前先清掉该区间，避免重复导入翻倍。
     *
     * @return array{file:string,company_id:int,from:?string,to:?string,orders:int,rows:int,no_label_orders:int,skipped:int,labels:int,seconds:float}
     */
    public function importFromFile(
        string $path,
        ?string $from = null,
        ?string $to = null,
        int $companyId = 1,
        bool $replace = false,
        string $source = 'jst-order-label-file-import',
    ): array {
        $from = $from === null || trim($from) === '' ? null : $this->normalizeDate(trim($from), '起始日期');
        $to = $to === null || trim($to) === '' ? null : $this->normalizeDate(trim($to), '结束日期');
        if ($from !== null && $to !== null && $from > $to) {
            throw new InvalidArgumentException('起始日期不能晚于结束日期。');
        }
        if ($replace && ($from === null || $to === null)) {
            throw new InvalidArgumentException('按文件导入并使用 --replace 时，必须同时给出 --from 与 --to，才能确定要清空的区间。');
        }

        $start = $from === null ? null : $from . ' 00:00:00';
        $end = $to === null ? null : date('Y-m-d', strtotime($to . ' +1 day')) . ' 00:00:00';

        $target = $this->databases->connection();
        $startedAt = microtime(true);

        // 清空区间推迟到第一次成功读到数据行之后：万一文件打不开或表头不对，
        // 不能先把线上已有行删掉又导入失败。
        $clearWindow = function () use ($target, $companyId, $start, $end): void {
            $target->table('jst_order_labels')
                ->where('company_id', $companyId)
                ->where('ordered_at', '>=', $start)
                ->where('ordered_at', '<', $end)
                ->delete();
            app(PlantingOrderLabelLookup::class)->forget($companyId);
        };
        $cleared = ! $replace;

        $now = now();
        $orders = 0;
        $rows = 0;
        $noLabelOrders = 0;
        $skipped = 0;
        $seen = [];
        $orderSeen = [];
        $buffer = [];

        $flush = function () use (&$buffer, &$rows, $target, $companyId): void {
            foreach (array_chunk($buffer, self::INSERT_CHUNK) as $chunk) {
                $target->table('jst_order_labels')->insert($chunk);
                app(PlantingOrderLabelLookup::class)->forget($companyId);
                $rows += count($chunk);
            }
            $buffer = [];
        };

        $reader = $this->exportReader ?? app(XlsxSheetReader::class);

        // 第一遍：挑出「有标签」的订单号。聚水潭导出是一单多行，同一单常常只有部分
        // 明细行带标签；不按订单判断就会给本来有标签的订单多写一行「（无标签）」。
        $labeledOrders = [];
        $anyRow = false;
        foreach ($reader->rows($path, self::EXPORT_COLUMNS, self::EXPORT_OPTIONAL_COLUMNS) as $line) {
            $anyRow = true;
            if ($this->labelsForRecord(
                (string) ($line['标签'] ?? ''),
                (string) ($line['卖家备注'] ?? ''),
            ) === []) {
                continue;
            }
            $orderNo = trim((string) ($line['线上订单号'] ?? ''));
            if ($orderNo !== '') {
                $labeledOrders[$orderNo] = true;
            }
        }
        if (! $anyRow) {
            throw new RuntimeException('订单导出文件里没有数据行：' . basename($path));
        }

        if (! $cleared) {
            $clearWindow();
            $cleared = true;
        }

        // 第二遍：按「线上订单号 + 标签」去重后写入。
        foreach ($reader->rows($path, self::EXPORT_COLUMNS, self::EXPORT_OPTIONAL_COLUMNS) as $line) {
            $orderNo = trim((string) ($line['线上订单号'] ?? ''));
            $orderedAt = $this->normalizeOrderedAt((string) ($line['下单时间'] ?? ''));
            if ($orderNo === '' || $orderedAt === null) {
                $skipped++;

                continue;
            }
            if ($start !== null && ($orderedAt < $start || $orderedAt >= $end)) {
                $skipped++;

                continue;
            }

            $labels = $this->labelsForRecord(
                (string) ($line['标签'] ?? ''),
                (string) ($line['卖家备注'] ?? ''),
            );
            $withoutLabel = $labels === [];
            if ($withoutLabel) {
                if (isset($labeledOrders[$orderNo])) {
                    continue;
                }
                $labels = [self::NO_LABEL];
            }

            if (! isset($orderSeen[$orderNo])) {
                $orderSeen[$orderNo] = true;
                $orders++;
            }

            $shopName = mb_substr(trim((string) ($line['店铺名称'] ?? '')), 0, 191);

            foreach ($labels as $label) {
                $fingerprint = hash('xxh3', $orderNo . "\0" . $label);
                if (isset($seen[$fingerprint])) {
                    continue;
                }
                $seen[$fingerprint] = true;
                if ($withoutLabel) {
                    $noLabelOrders++;
                }

                $buffer[] = [
                    'company_id' => $companyId,
                    'order_no' => mb_substr($orderNo, 0, 100),
                    'shop_name' => $shopName,
                    'ordered_at' => $orderedAt,
                    'label' => $label,
                    'created_from' => $source,
                    'created_at' => $now,
                    'updated_at' => $now,
                ];
            }

            if (count($buffer) >= self::INSERT_CHUNK * 2) {
                $flush();
            }
        }

        $flush();

        $labelQuery = $target->table('jst_order_labels')->where('company_id', $companyId);
        if ($start !== null) {
            $labelQuery->where('ordered_at', '>=', $start)->where('ordered_at', '<', $end);
        }
        $labels = $labelQuery->distinct()->count('label');

        app(JstOrderLabelOptions::class)->forget($companyId);
        app(JstOrderLabelOptions::class)->warm($companyId);

        return [
            'file' => basename($path),
            'company_id' => $companyId,
            'from' => $from,
            'to' => $to,
            'orders' => $orders,
            'rows' => $rows,
            'no_label_orders' => $noLabelOrders,
            'skipped' => $skipped,
            'labels' => (int) $labels,
            'seconds' => round(microtime(true) - $startedAt, 1),
        ];
    }

    /** 下单时间归一成 Y-m-d H:i:s；解析不了就返回 null（该行跳过）。 */
    private function normalizeOrderedAt(string $value): ?string
    {
        $value = trim($value);
        if ($value === '') {
            return null;
        }

        $timestamp = strtotime($value);

        return $timestamp === false ? null : date('Y-m-d H:i:s', $timestamp);
    }

    /** @return list<string> */
    private function splitLabels(string $labels): array
    {
        $out = [];
        foreach (explode(',', $labels) as $label) {
            $label = trim($label);
            if ($label !== '') {
                $out[] = mb_substr($label, 0, 191);
            }
        }

        return array_values(array_unique($out));
    }

    /** @return list<string> */
    private function labelsForRecord(string $labels, string $sellerRemark): array
    {
        $labels = $this->splitLabels($labels);
        if (stripos($sellerRemark, self::SELLER_REMARK_LABEL) !== false
            && ! in_array(self::SELLER_REMARK_LABEL, $labels, true)) {
            $labels[] = self::SELLER_REMARK_LABEL;
        }

        return $labels;
    }

    private function normalizeDate(string $value, string $label): string
    {
        $value = trim($value);
        if (preg_match('/^\d{4}-\d{2}-\d{2}$/', $value) !== 1) {
            throw new InvalidArgumentException($label . '必须是 YYYY-MM-DD。');
        }

        [$year, $month, $day] = array_map('intval', explode('-', $value));
        if (! checkdate($month, $day, $year)) {
            throw new InvalidArgumentException($label . '不是有效日期。');
        }

        return $value;
    }
}